Skip to content

Remove intel-pti==0.12.4 pin since we switched to DLE 2025.2 #5015

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
wants to merge 3 commits into
base: main
Choose a base branch
from
Draft

Remove intel-pti==0.12.4 pin since we switched to DLE 2025.2 #5015

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
470cba8
Remove intel-pti==0.12.4 pin since we switched to DLE 2025.2
anmyachev Sep 2, 2025
367ecad
Merge branch 'main' into amyachev/remove-pti-pin
anmyachev Sep 3, 2025
c98980d
Merge branch 'main' into amyachev/remove-pti-pin
anmyachev Sep 4, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 0 additions & 26 deletions .github/workflows/triton-benchmarks.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -126,16 +126,10 @@ jobs:
run: |
cd benchmarks
pip install .
pip install intel-pti==0.12.4
PTI_LIBS_DIR=$(python -c "import sysconfig; print(sysconfig.get_paths()['stdlib']+'/..')")
# the output should contain: `libpti.so`, `libpti_metrics.so.0.12.4` and `libpti_view.so.0.12.4`
ls $PTI_LIBS_DIR
echo "PTI_LIBS_DIR=$PTI_LIBS_DIR" >> $GITHUB_ENV

- name: Run Triton Softmax kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'fused_softmax.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'fused_softmax.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python fused_softmax.py --reports $REPORTS --n_runs $N_RUNS
source ../../scripts/capture-hw-details.sh
Expand All @@ -146,7 +140,6 @@ jobs:
- name: Run Triton GEMM kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'gemm_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'gemm_benchmark.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python gemm_benchmark.py --reports $REPORTS --n_runs $N_RUNS
mv $REPORTS/matmul-performance.csv $REPORTS/matmul-performance-base.csv
Expand All @@ -160,7 +153,6 @@ jobs:
- name: Run Triton GEMM kernel benchmark - with tensor of pointer
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'gemm_tensor_of_ptr_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'gemm_tensor_of_ptr_benchmark.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python gemm_tensor_of_ptr_benchmark.py --reports $REPORTS --n_runs $N_RUNS
source ../../scripts/capture-hw-details.sh
Expand All @@ -173,7 +165,6 @@ jobs:
- name: Run Triton GEMM kernel benchmark - with tensor descriptor
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'gemm_tensor_desc_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'gemm_tensor_desc_benchmark.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python gemm_tensor_desc_benchmark.py --reports $REPORTS --n_runs $N_RUNS
source ../../scripts/capture-hw-details.sh
Expand All @@ -186,7 +177,6 @@ jobs:
- name: Run Triton GEMM (A@B^t) kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'gemm_benchmark.py_abt')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'gemm_benchmark.py_abt') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
TRANSPOSE_B=1 python gemm_benchmark.py --reports $REPORTS --n_runs $N_RUNS
mv $REPORTS/matmul-performance.csv $REPORTS/matmul-performance-bt.csv
Expand All @@ -198,7 +188,6 @@ jobs:
- name: Run Triton GEMM (A^t@B) kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'gemm_benchmark.py_atb')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'gemm_benchmark.py_atb') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
TRANSPOSE_A=1 python gemm_benchmark.py --reports $REPORTS --n_runs $N_RUNS
mv $REPORTS/matmul-performance.csv $REPORTS/matmul-performance-at.csv
Expand All @@ -210,7 +199,6 @@ jobs:
- name: Run Triton GEMM (stream-k) kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'gemm_streamk_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'gemm_streamk_benchmark.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python gemm_streamk_benchmark.py --reports $REPORTS --n_runs $N_RUNS
source ../../scripts/capture-hw-details.sh
Expand All @@ -220,7 +208,6 @@ jobs:
- name: Run Triton GEMM (split-k) kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'gemm_splitk_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'gemm_splitk_benchmark.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python gemm_splitk_benchmark.py --reports $REPORTS --n_runs $N_RUNS
source ../../scripts/capture-hw-details.sh
Expand All @@ -230,7 +217,6 @@ jobs:
- name: Run Triton GEMM + PreOp (exp) kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'gemm_preop_exp_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'gemm_preop_exp_benchmark.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python gemm_preop_exp_benchmark.py --reports $REPORTS --n_runs $N_RUNS
source ../../scripts/capture-hw-details.sh
Expand All @@ -239,7 +225,6 @@ jobs:
- name: Run Triton GEMM + PostOp (Gelu) kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'gemm_postop_gelu_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'gemm_postop_gelu_benchmark.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python gemm_postop_gelu_benchmark.py --reports $REPORTS --n_runs $N_RUNS
source ../../scripts/capture-hw-details.sh
Expand All @@ -248,7 +233,6 @@ jobs:
- name: Run Triton GEMM + PostOp (add matrix) kernel benchmark bfloat16
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'gemm_postop_addmatrix_benchmark_bfloat16.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'gemm_postop_addmatrix_benchmark_bfloat16.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python gemm_postop_addmatrix_benchmark.py --reports $REPORTS --n_runs $N_RUNS
source ../../scripts/capture-hw-details.sh
Expand All @@ -258,7 +242,6 @@ jobs:
- name: Run Triton GEMM + PostOp (add matrix) kernel benchmark int8
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'gemm_postop_addmatrix_benchmark_int8.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'gemm_postop_addmatrix_benchmark_int8.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
INT8_ONLY=1 python gemm_postop_addmatrix_benchmark.py --reports $REPORTS --n_runs $N_RUNS
source ../../scripts/capture-hw-details.sh
Expand All @@ -268,7 +251,6 @@ jobs:
- name: Run Triton FA fwd kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flash_attention_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flash_attention_benchmark.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python flash_attention_benchmark.py --reports $REPORTS --n_runs $N_RUNS

Expand All @@ -279,7 +261,6 @@ jobs:
- name: Run Triton FA bwd kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flash_attention_bwd_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flash_attention_bwd_benchmark.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
FA_KERNEL_MODE="bwd" \
python flash_attention_benchmark.py --reports $REPORTS --n_runs $N_RUNS
Expand All @@ -292,7 +273,6 @@ jobs:
- name: Run Triton FA fwd kernel benchmark - with tensor descriptors
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flash_attention_tensor_desc_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flash_attention_tensor_desc_benchmark.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python flash_attention_tensor_desc_benchmark.py --reports $REPORTS --n_runs $N_RUNS
mv $REPORTS/attn-performance.csv $REPORTS/attn-tensor-desc-performance.csv
Expand All @@ -304,7 +284,6 @@ jobs:
- name: Run Triton FlexAttention Causal Mask fwd kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flex_attention_benchmark_causal_mask.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flex_attention_benchmark_causal_mask.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python flex_attention_benchmark_causal_mask.py --reports $REPORTS --n_runs $N_RUNS

Expand All @@ -315,7 +294,6 @@ jobs:
- name: Run Triton FlexAttention (batch_size=4) Causal Mask fwd kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flex_attention_benchmark_batch4-causal_mask.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flex_attention_benchmark_batch4-causal_mask.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
BATCH_SIZE=4 python flex_attention_benchmark_causal_mask.py --reports $REPORTS --n_runs $N_RUNS

Expand All @@ -326,7 +304,6 @@ jobs:
- name: Run Triton FlexAttention (batch_size=16) Causal Mask fwd kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flex_attention_benchmark_batch16-causal_mask.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flex_attention_benchmark_batch16-causal_mask.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
BATCH_SIZE=16 python flex_attention_benchmark_causal_mask.py --reports $REPORTS --n_runs $N_RUNS

Expand All @@ -337,7 +314,6 @@ jobs:
- name: Run Triton FlexAttention Custom Masks fwd kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flex_attention_benchmark_custom_masks.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flex_attention_benchmark_custom_masks.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python flex_attention_benchmark_custom_masks.py --reports $REPORTS --n_runs $N_RUNS

Expand All @@ -351,7 +327,6 @@ jobs:
- name: Run Prefix Sums kernel benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'prefix_sums.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'prefix_sums.py') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/triton_kernels_benchmark
python prefix_sums.py --reports $REPORTS --n_runs $N_RUNS
source ../../scripts/capture-hw-details.sh
Expand All @@ -360,7 +335,6 @@ jobs:
- name: Run micro benchmark
if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'micro_benchmarks.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'micro_benchmarks') }}
run: |
export LD_LIBRARY_PATH=$PTI_LIBS_DIR:$LD_LIBRARY_PATH
cd benchmarks/micro_benchmarks
python run_benchmarks.py --reports $REPORTS

Expand Down
Loading